Categorisation is an essential process in human cognition and language development. It has been proposed that childrenʼs ability to categorise objects into categories proceeds through a differentiation from broader to finer distinctions (Mandler, Bauer, & McDonough, 1991). It has also been claimed that children learn first to categorise typical exemplars prior to more atypical members of the same category (Markman, 1989; Mervis & Pani, 1980; Rosch & Mervis, 1975).
By studying numeral classifier acquisition we are able to examine how children learn to categorise and label objects in their environment using a constrained system or framework. The current study aimed at investigating how children categorise objects into Malay shape-based numeral classifier categories using a discrimination task. Specifically, it was aimed at examining if categorising pairs of numeral classifier exemplars that have a greater number of differing features, a strong contrast (two features differ: in dimensionality and rigidity, or in dimensionality and size) is an easier task than categorising those that differ in just one feature, a weak contrast (difference in dimensionality only). In addition, this study aimed at investigating if degree of typicality of numeral classifier exemplars has an effect on childrenʼs performance in categorising the objects. A categorisation discrimination task was conducted on 140 6- to 9-year-old Malay children using e-Prime. The objects for categorisation in this experiment were pictures of eight Malay shape-based numeral classifier exemplars, which were paired based on number of contrasts (strong vs. weak) of the numeral classifier categories they are members of, and degree of typicality of exemplars (typical vs. atypical). Children were required to select the object from the picture pair that matched the numeral classifier category. It was found that children categorised objects more readily when there was a strong than weak contrast between exemplars of the shape-based numeral classifiers. Typical exemplars were categorised more readily than atypical exemplars. There was no interaction effect between the number of contrasts and typicality, which suggests that there are separate mechanisms operating in the categorisation process. This supports a mixed representational model of categorisation.